Upload
truongthu
View
235
Download
0
Embed Size (px)
Citation preview
AVS CMP User Group Meeting“Advances in Ceria Slurries to Address
Challenges in Fabricating Next Generation
Devices”
Daniel S. Dickmann
4/16/15
AVS CMP User Group Meeting
“Advances in Ceria Slurries to Address Challenges in
Fabricating Next Generation Devices”
Outline
Introduction to Ferro
Historical Products
Rate Accelerants
Fast Oxide CMP Slurry
Path to Next Generations of STI and Fast Oxide Slurries
Ferro was founded in 1919 in Cleveland, Ohio. Porcelain enamel was the Company’s original business and it remains an important product line today.
We are a leading, global producer of performance materials and chemicals sold to manufacturers worldwide.
Our customers represent more than 30 industries and span 100-plus countries.
We have manufacturing facilities in The Americas, Europe/North Africa and Asia-Pacific.
Approximately 4,600 employees work at sites in 26 countries.
Ferro is publicly-owned; our shares trade on New York Stock Exchange under the symbol FOE.
We are a leading, global producer of performance materials and chemicals
3
Ferro’s Core Technologies
Particle Engineering
Color and Glass Science
Surface Chemistry and Surface Application
Formulation
Penn Yan Site
Located on 63 Acres
Lake Frontage of 1100 feet
320,000 Ft2 Under Roof
Global R&D Center: Dielectrics and
Surface Technologies
Customer Service Center for:
Cleveland, OH
Vista, California
Penn Yan, NY
Penn Yan Products• Manufacturer of ceramic powders and slurries
Dielectrics
• High purity engineered powders and formulations for the multi-layer
ceramic capacitor (MLCC) industry
Surface Finishing Material• Zirconia, Ceria and Alumina based powder and slurry formulations for
the LCD glass, glass ceramic hard disk, flat glass, cover glass,
precision glass, sapphire, plastic lens, metals and automotive
polishing applications
CMP• Ceria based powders and formulations used in semiconductor
applications
Ferro’s Historical Ceria STI Slurrys
Dmean
(nm)
D0
(nm)
LPC’s >
0.5 um
% Ceria HDP Removal
Rate (A/min.)
Nitride Removal
Rate (A/min)
1st Gen
STI
Slurry
145 450 500,000 4 2000 <20 A/min.
2nd Gen
STI
Slurry
145 450 400,000 4 2700 <20 A/min.
New Device Requirements
As devices get more complex, the demands (in addition to
the basic planarity requirement), on the CMP slurries that
enable their construction increase:
Defects
• Continuous push for reduction in defects
• LPC control has been demonstrated to be key
Rate
• More complex designs require:
– An increase in removal rate due to the number/thickness of layers requiring polishing
– The ability to polish multiple wafer/layer types with the same slurry
How do we get there?
Fundamental understanding is key!
Ceria Polishing Mechanism:
Surface Chemical Action
As opposed to other abrasive types (ie. Silica, alumina, zirconia, etc.), ceria
abrasives have a large surface chemical action during glass polishing
• Lee Cook, Journal of Non-Crystalline Solids 120 (1990), p. 152-171
By control of the ceria particle characteristics and the surface active chemical
components in a formulation, ceria systems can be highly tunable in terms of
oxide removal rates
Polishing Action
Ceria Polishing Mechanism:
Surface Oxidation State
H2O2 can shut down oxide removal rates using ceria slurries (more Ce4+ on
the surface shuts down oxide removal rates)
– Effect not seen with other abrasives (silica, zirconia, alumina, titania)
• R. Manivannan, S. Ramanathan, Applied Surface Science 255 (2009), 3764-3768.
Further studies have shown that Ce3+ sites on the surface of ceria particles is
critical for the silicon dioxide removal rate
• Veera Dandu (Clarkson thesis, also presented at 17th Annual International Symposium on Chemical
Mechanical Planarization, August 12th-15th, 2012, Lake Placid, NY)
Silica abrasive Ceria abrasive
Cerium Surface Oxidation State
Ferro has developed an additive that stabilizes the Ce3+ on the surface
of the ceria, leading to a higher population of Ce3+ sites and the
acceleration of the oxide removal rate
– Additive is stable in solution (beyond 12 month shelf life)
– Additive also buffers pH in low regime (pH=3-4)
Novel Rate Accelerant Chemistry
Particle145 nm
ceria
130 nm
ceria
130 nm
ceria
120 nm
ceria
pH low low high low
Accelerant
Packageno no no yes
All systems tested at 3 wt. % ceria
Extremely fast TOX removal rate!
Current HVM Fast Oxide Slurry
Fast Oxide Slurry was formulated with extra ceria and accelerant (+/+)
and reduced ceria and accelerant (-/-) and compared to standard
Formulation proven to be robust across formulation space studied
0
2000
4000
6000
8000
10000
12000
14000
16000
TOX Nitride Poly BPSG
Re
mo
val R
ate
(A
/min
)
Control
+/+
-/-
>15,000 Angstroms/minute
BPSG removal!
Ferro’s Ceria STI Slurry Evolution
Dmean
(nm)
D0
(nm)
LPC’s >
0.5 um
% Ceria HDP Removal
Rate (A/min.)
Nitride Removal
Rate (A/min)
1st Gen
STI
Slurry
145 450 500,000 4 2000 <20 A/min.
2nd Gen
STI
Slurry
145 450 450,000 4 2700 <20 A/min.
3rd Gen
STI
Slurry
130 300 <40,000 3 3500 <20 A/min.
4th Gen
STI
SRS-2092
130 300 <10,000 <0.5 2700 <20 A/min.
Continuous improvement toward lower ceria loading and lower
LPC’s for lower wafer level defectivity and lower Fab cost of
ownership
SRS-2092: Low Solids STI Slurry
Ferro has designed its next generation STI slurry
It is designed to be diluted (typically 1 part slurry : 2 parts water)
– Lower COO at <0.5% ceria loading
– Lower Defectivity at smaller particle sizes
0500
100015002000250030003500400045005000550060006500700075008000850090009500
100001050011000
-100 -75 -50 -25 0 25 50 75 100
Th
ick
ne
ss
(A
)
R(mm)
Initial & Residual Vs. Radius
Remaining Removed
0
500
1000
1500
2000
-100 -75 -50 -25 0 25 50 75 100
Th
ick
ne
ss
(A
)
R(mm)
Initial & Residual Vs. Radius
Remaining Removed
HDP
2658 A/min RR
Nitride
4 A/min RR
SRS-2092 Over-Polish Behavior
Extremely long overpolish window with excellent dishing performance!
Trench Oxide Values
Active Oxide Values
Endpoint Time based
upon Act. Ox. data
Ideal Trench Oxide Thickness = 5500A
Where to next?
Reduced LPC’s/smaller particles to further reduce
defects
Next Generation Fast Oxide Slurry
Dmean
(nm)
D0
(nm)
LPC’s > 0.5
um
TOX RR
(Ang./min)
TEOS RR
(Ang./min.)
BPSG RR
(Ang./min.)
Std. Fast
Oxide
130 296 100,000 7189 12,250 14,821
Low
Dmean
69 131 1,700 6139 11,651 11,352
70 nm ceria with extremely fast removal rates
130 nm ceria 70 nm ceria
Low Dmean SRS-2092
Ferro has developed a low Dmean version of SRS-2092 that has similar polish
behavior at approximately half the particle size
SRS-2092
~ 130 nm
LDM SRS-2092
~ 70 nm
Where to next?
Further removal rate improvements
How will we get there?
Ferro’s Rate Increase
Developmental Program
Ferro is in the middle of a large rate increase developmental
program
A fundamental 3 path approach is being taken:
Alternate Ceria’s
Alternate Processing
Alternate Accelerants
Current State
Based upon 120 nm Dmean
– With the intent that defects will be reduced compared to Std. Fast Oxide Slurry
– LPC’s are a similar order of magnitude as the sub 70 nm particles
Combination of alternate processing and alternate accelerants
– Increase in oxide removal rates with smaller particles (even with a reduction in ceria %)
Path Forward
Removal Rate Improvement:
– Optimize best options to date to achieve greater than 40% increase in removal rate vs. Std. Fast
Oxide Slurry with improved defect levels
– Continue fundamental work to lead to breakthroughs necessary to increase removal rates greater
than 80% vs. Std. Fast Oxide Slurry while further improving defect levels
Defect Improvement:
– Integrate the learnings above with 70 nm and smaller particles
– Incorporate the learnings above in next generation STI Slurries to enable the use of 70 nm and
smaller particles and further reduce ceria levels
Acknowledgements
Bob Her
Brian Santora
Nate Urban
Mohammed Megherhi
Mike Maxwell
Dave Walker